Mapping the 3D structure of almond trees using UAV acquired photogrammetric point clouds and object-based image analysis

• Published in: Biosystems engineering Vol. 176; pp. 172 - 184
• Main Authors: Torres-Sánchez, Jorge, de Castro, Ana I., Peña, José M., Jiménez-Brenes, Francisco M., Arquero, Octavio, Lovera, María, López-Granados, Francisca
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2018
• Subjects: Crown segmentation; Drone; Orchard architecture; Precision agriculture; Volume; Woody crop; Precision agriculture; Crown segmentation; Drone; Woody crop; Volume; Orchard architecture
• ISSN: 1537-5110; 1537-5129
• DOI: 10.1016/j.biosystemseng.2018.10.018

The geometric features, such as canopy area, tree height and crown volume, of agricultural trees provide useful information to elucidate plantation status and to design input prescription maps adjusted to real crop needs. This work presents an innovative procedure for computing the 3-dimensional (3D) geometric features of almond trees by applying two phases: 1) generation of photogrammetric point clouds with unmanned aerial vehicle (UAV) technology, and 2) analysis of the point clouds using object-based image analysis (OBIA) techniques. To test this approach, a UAV with a visible-RGB (low-cost) sensor was flown over three experimental almond groves at different phenological stages for two years, and the validation field method consisted of registering the height of a total of 325 trees in the two fields. The OBIA algorithm developed in this study achieved successful results: i) the individual and overall similarity measures between manually delineated and automatically detected almond tree crowns were above 0.9, and ii) the validation assessment conducted to estimate tree height from the UAV-derived algorithm produced an R2 = 0.94, an overall root mean square error (RMSE) of 0.39 m. The information derived from the OBIA algorithm was used for generating 3D maps for every tree volume and volume growth, which would be useful to understand the linkages between tree and crop management operations in the context of precision agriculture, with relevant agro-environmental implications. Our findings show that an RGB, low-cost sensor on-board a UAV provided dense 3D point clouds can be used for accurately characterising almond tree architecture. •UAV derived photogrammetric point clouds were generated in 3 almond orchards.•An automatic OBIA algorithm was developed for analyzing the point clouds.•Automatic almond crown delineation and height estimation achieved high accuracies.•Maps for almond tree volume and volume growth were generated.